Technical availability looks only at stops caused by equipment problems, and it ignores changeovers, waiting and speed. OEE counts every loss in scheduled time. So a line can read 95.8% technical availability and 73.5% OEE in the same week, and the term has no single standard formula, so check which one a report uses.
Two numbers with two owners
Technical availability answers a question for the maintenance team: when we needed the machine, was it technically fit to run? OEE answers one for the whole plant: how much good product came out of the scheduled time?
Both are useful, but they often sit in the same meeting with no one explaining the gap. Most of this page is about that gap.
There is more than one formula
We found no single definition that everyone uses. Here is who uses which, as far as we could read.
| Who | Version | Note |
|---|---|---|
| Evocon, a software vendor | (Planned time - technical stops) / Planned time | Changeovers, cleaning and breaks do not count against it |
| Reliability engineering, in common use | MTBF / (MTBF + MTTR) | Uses only running time and repair time |
| VDI 3423:2011-08, a German guideline | Paywalled | We read its scope only. It is meant to support talks between plants and machine suppliers. |
The word "technical" is not settled either. Some plants count a jam as technical, some call it operational, and some treat waiting for a technician as part of the repair.
A worked week
Example: a line is planned for 7,200 minutes in a week. It has 6 breakdowns adding up to 300 minutes, 540 minutes of changeovers, 240 minutes waiting for material and 120 minutes with no operator.
That is 1,200 minutes of stops, so it runs 6,000 minutes. It also runs at 90% performance and 98% quality.
MTBF here is 6,000 ÷ 6 = 1,000 minutes of running per failure, and MTTR is 300 ÷ 6 = 50 minutes. The MTBF and MTTR calculator does this for you, and its MTTR must include all the time to restore the machine, not just hands-on work.
The two technical versions differ by 0.6 points on the same data. So a change of formula can look like a change in the machine.
Where the gap goes
From 95.8% to 73.5% is 22.3 points, and none of it was logged as a technical failure. Here is the whole week, as shares of planned time.
Check the sum: 300 + 900 + 600 + 108 + 5,292 = 7,200. The speed loss is 6,000 x 10% = 600 minutes, and bad units take 2% of the remaining 5,400 minutes.
So maintenance could cut breakdowns to zero, and OEE would only rise to about 77%. With the other losses unchanged, that is 6,300 run minutes x 90% x 98% = 5,556.6 good minutes over 7,200, or 77.2%.
That is a fact to share, not a reason to blame anyone. Most of the loss in this example belongs to planning, supply and setup, and no maintenance plan will fix that.
Technical availability vs plain availability
OEE's own availability factor is the closer cousin. Both are about time, and neither looks at speed or scrap.
The difference is what counts as a stop. OEE availability counts them all, including changeovers and induced stops, and the three factors guide shows how each behaves.
| What happened | OEE availability | Technical availability, stops version |
|---|---|---|
| Bearing failure | Lowers it | Lowers it |
| Product changeover | Lowers it | No effect |
| Waiting for material | Lowers it | No effect |
| Upstream machine down, this one starved | Lowers it | No effect here, though it may count on the upstream machine |
| Running slowly | No effect | No effect |
| Making scrap | No effect | No effect |
The starved case is covered in blocked and starved. It is a common reason why a machine and its line tell different stories.
When each one misleads
Technical availability has one built-in weakness: it depends on how stops are coded. Move 100 of the 300 minutes from "breakdown" to "waiting for parts", and the stops version jumps to 97.2%.
OEE does not move in that case, because every stop counts either way. That makes it a useful cross-check on a rising technical figure.
OEE has the opposite weakness. It will not tell a maintenance team whether its own work is improving, because planning problems can bury a real gain.
How to use both without a fight
State the formula, and list which reason codes count as technical.
Both numbers should come from the same records, with no second maintenance list.
Maintenance owns technical availability, and operations owns OEE.
Show the loss bars, so the gap has names on it.
Clear downtime reason codes do most of this work. The downtime percentage guide helps when people quote stops in other ways.
Both numbers hold up only when each stop has a time and a reason. Fabrico takes stops from the machine, and operators tag the cause at the line, so one record can feed both views. See how Fabrico tracks stops.
What people ask next
What is the formula for technical availability?
A common form is (planned time - unplanned technical stops) ÷ planned time. Reliability engineers often use MTBF ÷ (MTBF + MTTR). No single standard version applies everywhere, so state which one you use.
What is a good technical availability?
We found no neutral published benchmark we could stand behind, so we give none here. Compare the machine with its own trend, and with any figure in its supply contract. For OEE targets, see how to set OEE targets.
Is technical availability the same as uptime?
Not usually. Uptime normally counts all stops or all calendar time, depending on who measures it, while technical availability counts equipment problems only. Our uptime guide sets out the sums.
Does a changeover count against technical availability?
In the version we quote from Evocon, it does not, because a changeover is not a technical problem. It does count against OEE availability. So a plant with many changeovers can show a wide gap between the two.
Why is our technical availability high when OEE is low?
Because most of your losses are probably not equipment failures. Look at changeovers, waiting and speed loss first, and build the picture with a downtime Pareto chart.
Fabrico is a light MES for OEE. It takes stops, speed and counts from PLCs or IoT sensors, with add-on AI cameras counting output where a machine has no usable signal, and operators tag the cause at the line.